We Tried To Outguess The FiveThirtyEight Forecast
Welcome to FiveThirtyEight’s politics chat. The transcript below has been lightly edited.
nrakich (Nathaniel Rakich, senior elections analyst): Back by popular demand (a phrase which here means “Alex peer-pressured me into it”), it’s the FiveThirtyEight politics chat!
There are somehow only three weeks left until Election Day, and the FiveThirtyEight forecast is furiously updating several times a day now. As of Tuesday, Oct. 18, at 1 p.m. Eastern, Democrats had a 63-in-100 chance of holding onto the Senate in the Deluxe version of our forecast, and Republicans had a 73-in-100 chance of flipping the House.
But we know better than some computer, right?! (Editor’s note: No, the whole point of the forecast is to eliminate the biases that inevitably arise when humans make predictions.) OK, fine, good point, editor. But there are things that the model doesn’t know that we do — like which races the national parties have given up on and which candidates are too extreme or otherwise embroiled in controversy.
So today, we’re going to play a game of “Are You Smarter Than The FiveThirtyEight Forecast?” I’m going to name a few important races and cite the FiveThirtyEight forecast’s odds in them. Then you guys tell me whether you think the forecast is too good for Democrats or too good for Republicans. (If you want to be a coward, you can also say you think the forecast is about right.) Everyone ready?
Want to know who will win the House? Watch these 4 districts | FiveThirtyEight
geoffrey.skelley (Geoffrey Skelley, senior elections analyst): I am ready to outthink you, IBM Blue/Watson/whatever your name is.
santul.nerkar (Santul Nerkar, editor): I am ready to lose!
alex (Alex Samuels, politics reporter): This chat will prove why I’m not an election forecaster. 😉
nrakich: OK, let’s start with the Arizona governor’s race. According to our Deluxe forecast, Republican former newscaster Kari Lake has a 55-in-100 chance of winning and Democratic Secretary of State Katie Hobbs has a 45-in-100 chance. What say you?
alex: This might be an unpopular opinion, but I’m honestly surprised this race isn’t more squarely in Lake’s column. Our forecast and polling average have Lake and Hobbs essentially tied, but it just seems like Lake is the more visible candidate — especially lately. There are so many stories out there now about how she’s a rising star in the GOP and could even be a Republican presidential candidate in the future. I’m not sure if “charisma” is the right term to use here, but her experience as a TV personality, plus the fact that she’s been in the news a lot recently (kinda like former President Donald Trump!), could make her a tough candidate to beat.
santul.nerkar: At the risk of sounding like a coward right off the bat, I think the forecast here sounds about right! As our colleague Kaleigh Rogers wrote last month, Lake has much better odds in the governor’s race than Republican venture capitalist Blake Masters does in the Senate race. Part of that comes down to the opponents each candidate is facing, but it may also suggest that openly denying the legitimacy of the 2020 election might not be a death knell for one’s candidacy. (Lake has been one of the leading proponents of the baseless idea that the 2020 election was stolen from Trump, while Masters has been more circumspect.)
Plus, there are other reasons to be more bullish on Lake than on Masters. She enjoys high name recognition due to two decades as a local TV newscaster, as Alex mentioned, while Hobbs has faced widespread scrutiny on the right, thanks to her up-close-and-personal role administering the 2020 election.
alex: To be clear, I don’t think the forecast is wildly underestimating Lake’s odds in this race, because it’ll probably be a close one. At least one recent poll has Hobbs’s and Lake’s favorability numbers tied at 39 percent (Lake’s favorability was net -3 points, while Hobbs’s was net +3 points), and another poll showed that just over half the state’s registered voters (51 percent) classified Lake’s views as “extreme.” I think what differentiates Lake is her visibility/personality — and that could resonate with voters!
geoffrey.skelley: I also tend to view our forecast as about right on this one. The polls have been close, and Hobbs has outraised Lake, but Lake has had more outside financial help (mostly from the Republican Governors Association). Lake is a combative, controversial candidate, but Hobbs isn’t a force in the same way Democratic Sen. Mark Kelly is in the Senate contest. For one thing, she’s had to answer for a scandal regarding the firing of a Black staff member who served under her when she was in the state Legislature.
nrakich: Hmmmm. You guys have all made compelling points! I’m going to say the forecast is too good for Republicans, though. I don’t think we need to overthink this one: Lake is a far-right candidate without prior political experience, and Hobbs is a statewide elected official. Plus, Kelly is a pretty good bet to win the Senate race, and as Geoffrey wrote last week, Senate and governor’s races just don’t go in opposite directions very often anymore. But I agree that the race is close — I think I would just personally switch the odds (so Hobbs at 55-in-100, Lake at 45-in-100).
OK, let’s move on to the Pennsylvania governor’s race. This one is pretty different: Our Deluxe forecast gives Democratic Attorney General Josh Shapiro a 96-in-100 chance of winning. Republican state Sen. Doug Mastriano has just a 4-in-100 chance. Thoughts?
geoffrey.skelley: In a state that has a partisan lean just slightly to the right of the country, I have a hard time viewing a Democratic candidate as a safe bet — which is what being clear of 95 percent suggests. So I would say this feels too confident in Democrats’ chances. Shapiro is clearly a strong favorite, but nearly a sure thing in a midterm year with a Democrat in the White House in a purple state? No.
alex: Pennsylvania is an interesting state because, yes, I think it’s possible that the polling here might overestimate Democrats in both the Senate and governor’s races. But I agree with our forecast — and the takeaways of other prognosticators — that Shapiro has a clear edge here. It’s pretty telling that national Republicans have practically abandoned Mastriano.
nrakich: OK, but to play devil’s advocate: Shapiro leads in the polling average by 11 percentage points. That’s a huge lead, and it would require an historic polling error to overcome. Maybe the race will tighten, but you guys don’t think this is safely Democratic?
alex: To be clear, I’m inclined to agree with our forecast because the GOP kind of, uh, fumbled in Pennsylvania. Neither Republican Senate nominee Mehmet Oz nor Mastriano are particularly well-liked, and that could dampen turnout among the GOP base. According to a September Fox News/Beacon Research/Shaw & Company Research poll, 34 percent of Oz supporters and 27 percent of Mastriano supporters said they had “some reservations” about their candidate. (Twenty percent of Democratic Lt. Gov John Fetterman’s supporters and 19 percent of Shapiro supporters said the same.)
geoffrey.skelley: Nathaniel, I guess I’m just focused on the “fundamentals” — the non-polling factors that affect an election, like partisanship and candidate quality. When the fundamentals and polls are this far apart, I get a little uneasy (more on this later). Now, it is true that they can diverge more in governors’ races than in races for federal office, as voters are not as reflexively partisan about gubernatorial races. That certainly is part of why the Pennsylvania gubernatorial contest is where it is. But the forecast still appears too strong for Democrats.
santul.nerkar: Yeah, I’m still going to have to say our forecast is too confident in Democrats’ chances. Along with what Geoffrey said about Pennsylvania’s rightward tilt relative to the country, I refer back to Inside Elections’s rating of the race as “Lean Democratic” — which is closer to what our forecast says about the Senate race in the state. And if you view 96 percent as translating to “safe” for Democrats, I find it hard to believe there would be such a large divergence between what the experts and fundamentals are saying and what the polling is saying. I’ll take the side of the former.
nrakich: Yeah. And actually, I secretly agree with you guys: I think our forecast is too good for Democrats here.
OK, let’s stay in the Keystone State but switch campaigns. Our Deluxe forecast for the Pennsylvania U.S. Senate race gives Fetterman a 70-in-100 chance, while Oz has a 30-in-100 chance. What do you guys think?
alex: It’s interesting that the race here is tightening. I think that’s a sign that we may be overstating Fetterman’s strength here, too. If I were to bet today, I’d say that Fetterman still has a nice edge since he’s never trailed Oz in public polling. But the GOP’s attacks on Fetterman and his health may have weakened him as a candidate.
nrakich: Yeah, I think the forecast is about right here. As Alex said earlier, Pennsylvania Republicans just nominated an especially weak slate of statewide candidates this year. Mastriano is a far-right election denier, and Oz is super unpopular thanks to his reputation for carpetbagging from New Jersey and a nasty Republican primary that dragged down his unfavorable ratings.
geoffrey.skelley: We have almost no fresh polling in this race. There have been just two polls conducted entirely in October, and while both showed Fetterman just barely ahead of Oz, the lack of data here makes me nervous, especially because we saw things tightening based on polls in late September. So I’m inclined to say our forecast is overstating the Democrats’ edge here too. Still, Oz’s favorability numbers are strikingly bad: In late September, Suffolk University/USA Today found that just 34 percent of likely voters had a favorable view of Oz, versus 51 percent who said they had an unfavorable view. By comparison, Fetterman was in the mid-40s for both favorability and unfavorability. That may have changed — give us new polls! — but for now I’d still call Fetterman the favorite.
nrakich: Oh, good catch about the lack of recent polling, Geoffrey.
santul.nerkar: I think 70-in-100 sounds about right for Fetterman’s odds. Back in September, when our polling average found Fetterman leading by nearly 11 (!) points, Democrats’ overall chances in the Senate had reached a peak of about 70-in-100. But as the national environment has tightened, so has this race. I think the national forces that have slightly tempered Democrats’ chances in the Senate — decreasing enthusiasm around issues like abortion, and increasing focus on the economy — have also largely applied to the Pennsylvania Senate race.
alex: I love how “our forecast is largely right” has been the takeaway of this chat so far! 😎
nrakich: In Math We Trust.
OK, moving on to … Nevada! Our Deluxe forecast says Republican former Attorney General Adam Laxalt has a 51-in-100 chance of winning the U.S. Senate race, while Democratic incumbent Catherine Cortez Masto is at 49-in-100. Where would you guys place your bets?
alex: This race is going to give Democrats heartburn on Election Day, I suspect. A few months ago, I would’ve said that Cortez Masto was in a slightly better position than, say, Sen. Raphael Warnock in Georgia. But this is why I’m not an election forecaster! Laxalt doesn’t have the same baggage as some of his fellow GOP candidates, like Oz and Herschel Walker of Georgia. So I think this race might come down to whether Democrats are still on the upswing come Election Day or if a Republican-friendly environment bounces back. The fact that Laxalt “leads” Cortez Masto by less than a point in our polling average means that this race is truly a toss-up, but given the conventional wisdom around midterm elections favoring the party not in the White House, I give Laxalt a slight edge.
santul.nerkar: I’m inclined to think that our forecast is slightly underrating Republicans’ chances in Nevada. As FiveThirtyEight editor-in-chief Nate Silver recently wrote, Nevada isn’t that blue, and midterm elections have often dealt Democrats big losses in the Silver State. Moreover, Nevada doesn’t fit the demographic profile of states in which Democrats have gained in recent years: It has relatively few Black voters, and it’s among the states with the lowest share of college-educated voters — a group with whom Democrats have struggled in recent cycles.
geoffrey.skelley: This one feels right to me. Cortez Masto could be the most endangered Democratic incumbent on the ballot: As a former elected official with campaign experience, Laxalt is a better candidate for Republicans than what they have in states like Arizona, Georgia and Pennsylvania. But Cortez Masto also outraised Laxalt almost three-to-one in the third quarter of 2022, and while the GOP has more outside spending, candidate committees get much cheaper rates for running ads on TV and radio, so Cortez Masto and Democrats collectively are likely getting more bang for their buck. Moreover, Nevada midterm polling has a history of underestimating Democrats — former Senate Majority Leader Harry Reid, Cortez Masto’s predecessor, famously trailed by 2 to 3 points heading into Election Day 2010, only to then win by nearly 6 points. That doesn’t mean it will happen again, but these factors make me inclined to not write off Cortez Masto.
nrakich: Yeah, I was intentionally waiting to weigh in on this one…
I think you’re right, Geoffrey. The vibe ~online~ seems to be that Laxalt is now the favorite, and I’m just not sure that’s true. The race is definitely super close, though, so I’m going to trust the forecast on this one.
alex: This race also might come down to which issues voters prioritize more (surprise, surprise!). Of course, following the Supreme Court’s ruling overturning Roe v. Wade, abortion access should theoretically be a motivating issue for Democrats in the state. But since abortion access is enshrined in Nevada’s state constitution, I wonder whether voters will prioritize other issues, such as the economy? In a state like Nevada, that’d certainly make sense — especially considering the impact that the COVID-19 pandemic had on the state’s tourism industry.
nrakich: How about another Senate race that’s been in the news a lot lately: Georgia? Our Deluxe forecast currently gives Warnock a 59-in-100 chance of winning, while Walker has a 41-in-100 chance.
geoffrey.skelley: Unlike Pennsylvania, Georgia has been polled a lot recently, and Warnock has consistently led in every recent poll not sponsored by Walker’s campaign. Warnock’s leads have been slight, though, so the forecast feels about right — generally speaking, a toss-up, but Warnock with the slimmest of edges. In fact, Warnock has made some narrow gains, as you recently wrote, Nathaniel, in the aftermath of allegations that Walker, who opposes abortion rights, paid for a former girlfriend’s abortion in 2009. (Walker denies the claims.)
Our forecast shows about a 1-in-5 chance of a runoff, too, which also feels right to me. I’m not making any plans for the first week of December. …
alex: I’m not convinced! Consider that, a few weeks ago, this was arguably the most contested Senate race in the country, and some forecasters predicted that a runoff was possible. I think Walker’s abortion controversy impacted voters’ perception of him (as evidenced by the polls) and that Warnock does have an edge here — but I’m also not yet convinced that he’ll win the race outright in November. Even though Warnock’s polling lead has widened, he still sits at just 48 percent in our average. Now, if he were at 52 or 54 percent, I might feel differently!
nrakich: Yeah, with the 2021 Georgia runoffs a big exception, Georgia runoffs have historically benefited Republican candidates.
alex: Really, Nathaniel? If the race goes to a runoff, I’d think Warnock would be favored because Walker wouldn’t be able to ride Republican Gov. Brian Kemp’s coattails to victory. (Kemp is heavily favored in his race and has cleared 50 percent in some recent polls.)
nrakich: I mean, it’s very possible. Especially if control of the Senate is on the line again. I think that motivated Democrats to turn out and buck the trend in 2021.
geoffrey.skelley: Yeah, if we’re unlucky — election watchers like a break, too — we’ll get to find out whether 2021 is the new normal or an outlier. I can make the case that Warnock will benefit (what Alex said) or that Walker will (the pre-2021 runoff history).
alex: Siding with Alex over this supposed “history” sounds like a safer bet, Geoff! 🤪
geoffrey.skelley: I’m all about that history, though, Alex. But I remain open to new trends!
santul.nerkar: Something I’ve been tracking is whether Walker’s election odds ever dip below his alma mater’s chances of winning the College Football Playoff, according to FiveThirtyEight’s forecast. (As of now, that seems unlikely — Walker’s Senate chances are 41-in-100, while the Bulldogs have a 23 percent chance of defending their title.)1
On a more serious note, I think our forecast is properly appraising Walker’s chances. Yes, his polling has been shaky, especially after the abortion story. But Georgia still leans to the right of the nation as a whole, and the fundamentals suggest a closer race than the polling — which, again, has been consistently in Warnock’s favor. And to Nathaniel’s point, if this race goes to a runoff, history would seem to be on the side of Republicans.
Adding all of that together, I think an outlook that gives Warnock decent-but-not-overwhelming odds seems like a fair landing spot.
nrakich: I think you guys are probably making the more sensible calls, but I’m going to be a little 🌶 spicy 🌶 here and say that the forecast is too good for Republicans.
My hack for this game has been to look at the Classic version of the forecast, instead of the default Deluxe version. The Classic version blends the polls and fundamentals, while Deluxe adds in expert ratings like those from Inside Elections and Sabato’s Crystal Ball. And while Warnock is at 59-in-100 in the Deluxe version, he is at 68-in-100 in the Classic version.
That means the outside experts are more bullish on Walker than our forecast is by itself. And that … doesn’t really make sense to me. Walker is an incredibly flawed candidate — another political novice like Lake, and someone with multiple scandals and controversies surrounding him.
Given the rate at which bad headlines have drip, drip, dripped out about him, I wouldn’t be surprised if even more scandals emerge in these last three weeks of campaigning.
I assume the expert raters are being cautious with Georgia because of its history as a red state, but partisanship is accounted for in the forecast. Walker’s weakness as a candidate, IMO, is not (at least not fully).
alex: Spicy take, Nathaniel! I guess I still feel like it’s entirely possible that Kemp drags Walker across the finish line on Election Day. That’s another reason why it’s hard for me to say with certainty that Walker’s finished here!
geoffrey.skelley: Yeah, Alex, the gap between the margins in Georgia’s Senate and gubernatorial races has been pretty small in the past. But it looks like we’re going to get a larger one this time around, with Kemp leading by almost 6 points and Warnock by almost 4 points, per our polling averages. But how large will matter a great deal, because the more split-ticket voters there are, the better it is for Warnock, given Kemp’s stronger position.
santul.nerkar: I guess my question is how much of a candidate-specific “tax” or penalty Georgia voters will apply to Walker. To Alex’s point, Kemp seems to be on fairly solid footing in his race against former state House Minority Leader Stacey Abrams, while Georgia Secretary of State Brad Raffensperger also looks safe against his Democratic challenger. In other words, can a GOP Senate candidate run several points behind other statewide candidates and lose in a “red wave” year in Georgia? I’m not sold on that yet.
nrakich: If a red wave were developing, Santul, I’d agree with you! But so far, it looks like this will be closer to a neutral year than a 2010-style red wave.
Anyway, we can save that debate for our last question. 🙂
For now, one last Senate race. It’s not one of the top-tier ones, but I think it’s an interesting one for the purposes of this exercise: Ohio. The default, Deluxe version of our forecast gives Republican author J.D. Vance a 73-in-100 chance of winning and Democratic Rep. Tim Ryan a 27-in-100 chance. However, the polling is much tighter, and the Lite version of our forecast (which uses only polls) gives Vance only a 56-in-100 chance. Where would you guys peg the race?
geoffrey.skelley: Vance seems like a pretty mediocre candidate, as he’s been massively outraised and has struggled to consolidate support among the GOP base. However, I’ll be pretty stunned if he doesn’t win on Nov. 8, so count me as one who views our forecast as underselling the GOP’s odds. With an unpopular Democrat in the White House, I have a hard time seeing a Democrat winning a Senate race in a state that Trump won by 8 points. If it were 2018 (a good year for Democrats) and we had this matchup, then I’d be singing a different, toss-up tune. But it’s 2022, so I anticipate Vance will be in the Senate come January.
alex: I agree with the sentiment that Vance has an edge here. While I also don’t think that he’s a particularly strong candidate, it’ll be hard for Ryan to cut through in a light-red state. The fact that national Republicans have spent a ton of money on Vance’s behalf in the lead-up to Election Day signals to me that this will likely be a close contest, as the Lite version of our forecast suggests. But I think the fact that we’re even talking about a competitive Senate race in such a pro-Trump state speaks volumes about either Ryan’s strength as a candidate and/or Vance’s weakness as one.
santul.nerkar: Yeah, I don’t see a big reason to fault our forecast here, either. To keep things simple, it’s hard to imagine that just two years after Democrats strongly underperformed expectations in the presidential election — and where Biden’s approval is deep underwater — a Democrat will win a Senate seat. There are reasons to be skeptical of Vance, as he has attracted a lot of scrutiny for his about-face on Trump — going from the “Never Trump” camp of the GOP to a staunchly pro-Trump one. That being said, Vance hasn’t embroiled himself in scandal and controversy quite like, say, Walker, and it’s less clear whether Ohio voters will dock Vance for his transgressions.
nrakich: I’m with Geoffrey here — I think our Ohio forecast, even the Deluxe version, is still too good for Democrats. One tidbit that I think is overlooked: Ryan doesn’t actually have a history of overperforming with white, working-class voters the way his reputation suggests. In 2020, he did just 1.5 points better than Biden in his Youngstown-area congressional district.
OK, time for our last question! There are too many House races to talk about them individually, so we’ll just look at the lower chamber as a whole. As a refresher, our forecast gives Republicans a 73-in-100 chance of flipping the House, while Democrats have just a 27-in-100 chance of keeping control. What are your thoughts?
geoffrey.skelley: If you go district by district, I think it’s hard not to come up with an outcome in which the GOP nets at least five House seats — the mark they need to claim the House, which Democrats currently hold 222 to 213 — so I guess I’m a bit more bullish on their chances than the forecast is. For instance, the GOP should pick up five or six seats based on partisanship alone (with an assist from redistricting, which made several seats notably redder in redistricting). And history suggests that Republicans will eat into the toss-up seats held by Democrats, too. So I think the narrow margins here make a Republican majority somewhat more likely than roughly 3-in-4.
alex: I mean, I don’t think anyone’s writing off the very real possibility of a red wave. I’m inclined (once again!) to agree with our forecast’s odds of Republicans’ chances here. I know I jokingly bashed “history” before, but I just can’t see Democrats overcoming the conventional wisdom regarding midterm elections. …
nrakich: Agreed. The average midterm election since the end of World War II has seen the president’s party lose 26 House seats. And Democrats did a good job bucking that trend for much of the summer (thanks to the Supreme Court’s Dobbs decision), but polls of the generic congressional ballot are starting to move back toward Republicans. And while there’s no guarantee that that trend will continue, it does fit with the historical pattern of polls getting steadily worse for the president’s party as the midterms approach.
It should be noted that that is already baked into the forecast, which already expects the national environment to improve for Republicans. But the forecast, too, has ticked toward the GOP in the last couple weeks.
geoffrey.skelley: Yeah, the way I see it, I would give Republicans a better than 3-in-4 chance of getting the minimum they need to control the chamber. But in terms of the magnitude of their gains, I’m far less certain. According to the forecast, 80 percent of outcomes fall between the Republicans winning 210 seats (meaning there’d be a Democratic-controlled House) to 244 seats (a very strong Republican showing).
santul.nerkar: Yeah, I think the conventional wisdom would suggest that the House is still Republicans’ to lose in November — even with the bounce Democrats enjoyed over the summer after the overturning of Roe v. Wade. The last midterm in which a party that had the presidency went on to gain seats in the House was 2002, the first election after the Sept. 11 attacks and amid strong approval ratings for former President George W. Bush. Though there have already been watershed moments for the Biden administration that could serve to boost overall turnout — including the forgiveness of some student loans, the passage of the Inflation Reduction Act and the call to reschedule marijuana — Biden’s approval remains around 10 points underwater. That’s not a recipe for keeping the House, and our model reflects that reality.
But don’t y’all know? The House always wins.